Allow create_index_paths() to consider multiple join bitmapscan paths.

tglsfdc · tglsfdc · commit f5983923d81d · 2012-08-16T13:03:54.000-04:00
In the initial cut at the "parameterized paths" feature, I'd simplified
create_index_paths() to the point where it would only generate a single
parameterized bitmap path per relation.  Experimentation with an example
supplied by Josh Berkus convinces me that that's not good enough: we really
need to consider a bitmap path for each possible outer relation.  Otherwise
we have regressions relative to pre-9.2 versions, in which the planner
picks a plain indexscan where it should have used a bitmap scan in queries
involving three or more tables.  Indeed, after fixing this, several queries
in the regression tests show improved plans as a result of using bitmap not
plain indexscans.
diff --git a/src/backend/optimizer/path/indxpath.c b/src/backend/optimizer/path/indxpath.c
@@ -309,26 +309,92 @@ create_index_paths(PlannerInfo *root, RelOptInfo *rel)
 	}
 
 	/*
-	 * Likewise, if we found anything usable, generate a BitmapHeapPath for
-	 * the most promising combination of join bitmap index paths.  Note there
-	 * will be only one such path no matter how many join clauses are
-	 * available.  (XXX is that good enough, or do we need to consider even
-	 * more paths for different subsets of possible join partners?	Also,
-	 * should we add in restriction bitmap paths as well?)
+	 * Likewise, if we found anything usable, generate BitmapHeapPaths for the
+	 * most promising combinations of join bitmap index paths.	Our strategy
+	 * is to generate one such path for each distinct parameterization seen
+	 * among the available bitmap index paths.	This may look pretty
+	 * expensive, but usually there won't be very many distinct
+	 * parameterizations.
 	 */
 	if (bitjoinpaths != NIL)
 	{
-		Path	   *bitmapqual;
-		Relids		required_outer;
-		double		loop_count;
-		BitmapHeapPath *bpath;
+		List	   *path_outer;
+		List	   *all_path_outers;
+		ListCell   *lc;
 
-		bitmapqual = choose_bitmap_and(root, rel, bitjoinpaths);
-		required_outer = get_bitmap_tree_required_outer(bitmapqual);
-		loop_count = get_loop_count(root, required_outer);
-		bpath = create_bitmap_heap_path(root, rel, bitmapqual,
-										required_outer, loop_count);
-		add_path(rel, (Path *) bpath);
+		/*
+		 * path_outer holds the parameterization of each path in bitjoinpaths
+		 * (to save recalculating that several times), while all_path_outers
+		 * holds all distinct parameterization sets.
+		 */
+		path_outer = all_path_outers = NIL;
+		foreach(lc, bitjoinpaths)
+		{
+			Path	   *path = (Path *) lfirst(lc);
+			Relids		required_outer;
+			bool		found = false;
+			ListCell   *lco;
+
+			required_outer = get_bitmap_tree_required_outer(path);
+			path_outer = lappend(path_outer, required_outer);
+
+			/* Have we already seen this param set? */
+			foreach(lco, all_path_outers)
+			{
+				Relids		existing_outers = (Relids) lfirst(lco);
+
+				if (bms_equal(existing_outers, required_outer))
+				{
+					found = true;
+					break;
+				}
+			}
+			if (!found)
+			{
+				/* No, so add it to all_path_outers */
+				all_path_outers = lappend(all_path_outers, required_outer);
+			}
+		}
+
+		/* Now, for each distinct parameterization set ... */
+		foreach(lc, all_path_outers)
+		{
+			Relids		max_outers = (Relids) lfirst(lc);
+			List	   *this_path_set;
+			Path	   *bitmapqual;
+			Relids		required_outer;
+			double		loop_count;
+			BitmapHeapPath *bpath;
+			ListCell   *lcp;
+			ListCell   *lco;
+
+			/* Identify all the bitmap join paths needing no more than that */
+			this_path_set = NIL;
+			forboth(lcp, bitjoinpaths, lco, path_outer)
+			{
+				Path	   *path = (Path *) lfirst(lcp);
+				Relids		p_outers = (Relids) lfirst(lco);
+
+				if (bms_is_subset(p_outers, max_outers))
+					this_path_set = lappend(this_path_set, path);
+			}
+
+			/*
+			 * Add in restriction bitmap paths, since they can be used
+			 * together with any join paths.
+			 */
+			this_path_set = list_concat(this_path_set, bitindexpaths);
+
+			/* Select best AND combination for this parameterization */
+			bitmapqual = choose_bitmap_and(root, rel, this_path_set);
+
+			/* And push that path into the mix */
+			required_outer = get_bitmap_tree_required_outer(bitmapqual);
+			loop_count = get_loop_count(root, required_outer);
+			bpath = create_bitmap_heap_path(root, rel, bitmapqual,
+											required_outer, loop_count);
+			add_path(rel, (Path *) bpath);
+		}
 	}
 }
 
diff --git a/src/test/regress/expected/join.out b/src/test/regress/expected/join.out
@@ -2725,11 +2725,13 @@ where t1.unique1 = 1;
          Index Cond: (unique1 = 1)
    ->  Nested Loop
          Join Filter: (t1.ten = t3.ten)
-         ->  Index Scan using tenk1_hundred on tenk1 t2
-               Index Cond: (t1.hundred = hundred)
+         ->  Bitmap Heap Scan on tenk1 t2
+               Recheck Cond: (t1.hundred = hundred)
+               ->  Bitmap Index Scan on tenk1_hundred
+                     Index Cond: (t1.hundred = hundred)
          ->  Index Scan using tenk1_unique2 on tenk1 t3
                Index Cond: (unique2 = t2.thousand)
-(9 rows)
+(11 rows)
 
 explain (costs off)
 select * from tenk1 t1 left join
@@ -2743,32 +2745,36 @@ where t1.unique1 = 1;
          Index Cond: (unique1 = 1)
    ->  Nested Loop
          Join Filter: ((t1.ten + t2.ten) = t3.ten)
-         ->  Index Scan using tenk1_hundred on tenk1 t2
-               Index Cond: (t1.hundred = hundred)
+         ->  Bitmap Heap Scan on tenk1 t2
+               Recheck Cond: (t1.hundred = hundred)
+               ->  Bitmap Index Scan on tenk1_hundred
+                     Index Cond: (t1.hundred = hundred)
          ->  Index Scan using tenk1_unique2 on tenk1 t3
                Index Cond: (unique2 = t2.thousand)
-(9 rows)
+(11 rows)
 
 explain (costs off)
 select count(*) from
   tenk1 a join tenk1 b on a.unique1 = b.unique2
   left join tenk1 c on a.unique2 = b.unique1 and c.thousand = a.thousand
   join int4_tbl on b.thousand = f1;
-                                QUERY PLAN                                
---------------------------------------------------------------------------
+                               QUERY PLAN                                
+-------------------------------------------------------------------------
  Aggregate
    ->  Nested Loop Left Join
          Join Filter: (a.unique2 = b.unique1)
          ->  Nested Loop
                ->  Nested Loop
                      ->  Seq Scan on int4_tbl
-                     ->  Index Scan using tenk1_thous_tenthous on tenk1 b
-                           Index Cond: (thousand = int4_tbl.f1)
+                     ->  Bitmap Heap Scan on tenk1 b
+                           Recheck Cond: (thousand = int4_tbl.f1)
+                           ->  Bitmap Index Scan on tenk1_thous_tenthous
+                                 Index Cond: (thousand = int4_tbl.f1)
                ->  Index Scan using tenk1_unique1 on tenk1 a
                      Index Cond: (unique1 = b.unique2)
          ->  Index Only Scan using tenk1_thous_tenthous on tenk1 c
                Index Cond: (thousand = a.thousand)
-(12 rows)
+(14 rows)
 
 select count(*) from
   tenk1 a join tenk1 b on a.unique1 = b.unique2
